ANIMAL PRODUCTION. 271 



how ;i herd may be improved by the use of pure-bred sires. During the lirst 

 5 years the average yield of milli per cow per annum was 4,270 lbs., and the 

 average yield of butter fat 22.").? lbs. Corresponding figures for the past 2 

 years are 5,731 lbs. of millv and 272..'i lbs. of fat. 



"The complete records of all daughters which completed one or more years' 

 record are submitted for public record and study by those interested in the 

 problem, not because the records are phenomenal. The records are such as 

 any good farmer should duplicate, and are suggestive when it is recalled that 

 the average cow of the State produces less than 160 lbs. of fat per year." 



The records show a marked difference in the influence of the different sires 

 that have been at the head of the herd. 



Methods of breeding for the improvement of Michigan cattle. Organiza- 

 tion of cooperative cattle breeders' associations in Michigan, R. S. Shaw and 

 AV. F. Raven {Michigan ISta. Circ. Ji, in). 33-'iJf, figs. 3). — This cii'cular outlines 

 plans for improving the cattle of Michigan and contains a report of the field 

 agent on the work done toward organizing cooperative cattle breeders' associa- 

 tions. Articles of agreement and by-laws suggested for such associations are 

 included. 



The sheep stock of Gloucestershire, B. Swanwick and J. T. Hobbs {Jour. 

 Roy. Agi: 8oc. England, 69 {1908), pp. 32-4S).— This is a short account of 

 sheep breeding in the county of Gloucestershire, with special reference to the 

 development of the Cotswold and Oxford Down grades of sheep. 



A biometrical study of e^s production in the domestic fowl. I, Variation 

 in annual e^g production, R. Pearl and F. M. Surface ( U. 8. Dept. Agr., Bur. 

 Auim. Indus, Bui. 110, pt. 1, pp. 80, figs. 17). — This is the first of a series 

 dealing with a biometrical study of the results obtained at the Maine Station 

 in cooperation with the Bureau of Animal Industry of this Department in breed- 

 ing for egg production since the introduction of the use of trap nests in 1898. 

 The plan of the breeding work has been previously noted (E. S. R., 18, p. 471). 



Some of the problems discussed are variation curves of ovulation statistics 

 and their biometric constants, relation of variation to fertility, influence of 

 environment on variation constants, influence of selection on egg production, 

 and relation of first year to second year production. The breeds used in the 

 M-ork reported were Barred Plymouth Rocks and White Wyandottes, all from 

 homogeneous stock. A small number of Light Brahmas were used at first but 

 were soon discarded. The first year production was considered to be a good 

 measure of total production, and the few birds which laid no eggs the first 

 year have been excluded from the statistical analysis. The year 1902-3 was 

 not regarded as a normal year for the Plymouth Rocks, as they were laying in 

 August and molted in December. The years 1903-4 and 1906-7 include records 

 of 11 months only. 



In grouping the frequency distribution the ordinary methods of biometric 

 analysis were used. A class unit of 15 eggs was adopted. In deducing the 

 moments from the grouped material Sheppard's corrections were used. In 

 deducing constants from the ungrouped material the mean and median were 

 calculated by ordinary arithmetical methods. 



In getting curves to graduate statistics raw moments were found to be less 

 satisfactory than corrected moments. The units used were large but there was 

 a fairly close approximation at both ends of the range. In calculating the 

 standard deviation the method of moments was used. Other fornnihe used are 

 discussed in detail. The authors summarize their findings as follows: 



"Variation in annual egg production exhibits the following characteristics: 

 (a) The ol)served range of variation is from zero to about 250 eggs; {b) the 

 3406— No. 3—09 G 



